In this video I demonstrate how the power of stats and form can be used to improve the selection criteria and the profits in automated betting on horse racing.
The video ended up being a bit longer than I anticipated, and there’s still more to cover in a future video or two, so to help you navigate it, chapters have been added accordingly.
00:00 Introduction
01:49 Recent results
05:13 Configuring stats filters in ANZ Horse Bot
06:24 The stats the bot is using
07:20 The stats engine: ANZ Horse Ratings
11:35 Using the ANZ Ratings Analyser to hone strategies
19:20 Configuring the best performing strategy in the Analyser and Bot
20:20 Building and testing new strategies in the Analyser
25:22 Exploring results in the analyser
27:20 Some high-priced winners from Saturday
28:24 How the bot used the stats to find the winner of Belmont race 4
31:10 Using the analyser to check your strategies are performing as they should
33:38 The settings for the 2nd best strategy
36:01 The first winner for the day pays $8.40
36:51 Summary
Wrapping up the series on finding value by hedging and dutching greyhounds in Hedge Dog Bot, Samuel looks at the results of the various methods tested in recent weeks and highlights the ones which show the most potential for profit.
Continuing from last week’s video of finding value on the dogs by dutching at level stakes, this week I provide an update on results from last week’s strategy, and continue looking for more value at level staking by adding a hedge bet (insurance) on the favourite, with results from some days of that. Plus, the next stage in testing, and some exciting news about the return of British greyhound racing to free to air television.
As a number of people who watch these videos have expressed an interest in level staking, as opposed to the progression or recovery staking systems in most of the videos to date, I have set about trying to find an automated system which can deliver profit with level stakes on the greyhounds. The trick is to find odds which exceed the strike rate of certain runners.
In smaller greyhound fields, the outsiders of the field tend to present this value while still having a reasonable chance of winning races and delivering a profit.
My testing hasn’t all been smooth sailing but it has led me to a system which I’m now testing and has started off well. Here’s a look at what I’ve tested and what I’m doing now. And in a week I’ll come back with more results.
I received an email the other day with some questions about obtaining data from the betting bots I’ve been using and demonstrating, and graphing that data.
For the most part I think it’s a fairly straight forward process, but there’s a few little tips and tricks which can make the data more useful, so I thought it warranted a video.
With a 57% strike rate this is a decent strategy, but this is not a “set and forget” system. In order to be profitable it does require a few minutes of management each day.
This is using the Stop At A Winner Greyhound Deluxe software to dutch bet (proportionate bet sizes per runner to receive the same amount of profit regardless of which one of them wins) the top two favourites in the market. In my testing over the course of two weeks, it has generally returned a daily profit of about 100x whatever the per-winning-race profit target is set to.
There have been some drawdowns and stop losses hit along the way, which is why a bit of daily management is required to keep it on track. I explain the details of how to do this in the video, along with a second way of managing it which returns a lower profit but also carries a lower risk of hitting a stop loss and so potentially could be a better way to manage the system in the long run.
Thank you to all of you for your patience during my recent absence. I would like to especially thank those of you who offered your kind wishes to my mother while she was in hospital. Mum is back at home now, happy and comfortable, but still requires surgery at a date in the future which is yet to be set.
Today I have for you a rather simple little strategy which works quite slowly but also quite steadily on the horse racing markets. This is looking at the place market and looking to lay the favourite if it is at odds of 1.60 to 1.99. This generally indicates that although it is the favourite, it is quite a weak favourite, and often this results in it not running a place. Most favourites tend to sit at 1.50 or lower on the place market so this strategy doesn’t get a lot of bets but it does tend to get a good percentage of successful bets.
I am running this on races globally, wherever Betfair offers a place market. This is generally every horse and harness race in Australia and New Zealand, just about every race in the UK and Ireland, and then some of the French, South African and American races. The settings are fairly straight forward.
(click image to enlarge)
On these settings I would recommend a balance of $200 as a starting point (and yes, I know my balance is less than that in the screenshot, but I had some unexpected expenses recently and withdrawing some funds from Betfair was the best option at the time, and it does mean I’m running a bit of a risk having less than an ideal amount of funds to work with for this strategy but it has been steady over the month that it has been running so I’m willing to take that risk at this point in time).
These graphs show you results over the last month of testing, running a Stop At A Winner Staking method. You can see it has made approximately $80 in the month. Not a huge amount, but it has done so in a fairly steady manner with only two brief drawdowns of note.
It’s worth noting that the place market doesn’t have a heap of liquidity and not every race globally has a place market, and the number of races with a favourite in the target range is a fairly small percentage of events. But it is being selective in this manner which allows the system to flourish. Keeping the odds under 2.00 also means the stop at a winner staking method escalates in bet size quite slowly, which also helps to maintain it as a fairly safe and steady system. And the Stop At A Winner Deluxe software handles the job quite nicely.
A few people have been asking when I would have a soccer strategy ready to publish, well here it is.
I fiddled with it for a little while before arriving at these settings, which have now been in place for two weeks, and I’m pleased to say there has been some fairly steady profit.
For the first week or so I had it running on a 50c base stake and it produced a profit of about $16.
For the second week when I had more confidence in it, I increased the base stake to $2.50 in a balance of $100. It generated $50 of profit. Impressively there was only one loss in this period.
If we put both periods together and look at how running at a $2.50 base stake over the entire period would have looked, it would have generated $130.
This is using the LayPro88 Football Bot and is placing lay bets on the favourite in correct score market, if the home team is heavily favoured to win and the favourite in the correct score market is paying between $5.00 and $7.80, as long as the favourite score is not 1-1 (which is unlikely anyway if the home team is strong favourite to win). Full details of this can be seen in the video, along with how I’m managing multiple tabs to boost the potential profits from the strategy. However it would also work with just one tab running, perhaps with lower profit, but also lower risk, and so I have details on how I would suggest setting that up in the video too.
Hopefully this has been worth the wait. I know a few people have been eager to see this released.
As is probably obvious by now, I am using a few betting bots produced by Steve and Michael. Over time using them I have developed a few habits and routines to keep them running at their best and most efficient. I thought it was worth creating a bit of a guide on some of the things I do to keep the bots running optimally.
The video covers everything from preventing Windows Update from interrupting 24/7 bot operations until I’m ready for them to run, managing memory usage to keep bots running quickly, using the tabs in ways which enhance profitability, and keeping archives of results without needing to keep them loaded in the bots, and some other bits and pieces which I find helpful.
The Windows Update Blocker utility which I’m using to give me complete control over when Windows Updates get to run is a free utility from Sordum. I cannot emphasise enough that you should NOT use this to prevent Windows Updates from running completely, but rather only to prevent them from running at inconvenient times. Windows Updates are an important function on any Windows-based computer for maintaining security.
If you’re comfortable using Windows Group Policy or the Windows Registry Editor and you have a professional version of Windows 10 or 11, or a server edition of Windows, Windows Central has a guide on how to manage Windows Updates through those methods without needing a third-party utility. This is probably a better method than the Windows Update Blocker utility, but it doesn’t work on home editions of Windows and I have seen Windows ignore it occasionally if it thinks the update it wants to install is particularly important.
And the bots I’m currently running and are featured in the video, although my hints and tips apply more broadly to more bots than just these.
I’ve been quite pleased with the results of the horse strategies which I ran through last week, and the greyhounds have been doing well too, but required more patience than I could provide so I made a change the other day and it’s been quite pleasing. This morning I decided it was a good time to update you on the results.
Shyley insisted on having her say around the seven minute mark and Pebbles joined her for a cameo.
I also gave a sneak peek of a soccer strategy I’m working on. I had hoped to provide details by now but I made some big changes to the strategy the other day and it hasn’t been running for long enough (and I don’t have the data to run a decent simulation on) to have enough confidence in it to be willing to properly details the strategy yet. But it’s started well so hopefully by early next week I can do that.
Some of the graphs in the video might be a touch small, so here they are. Click any of them to enlarge them.
Backing horses in the $2-$3 odds range in the Fav Money Maker Bot using a 50c starting stake, $25 profit target and $50 stop loss
It hit a stop loss early on as Friday wasn’t a great day for it (and my laying results for Saturday indicate Saturday would have been disastrous so it’s a good thing I followed my strategy and stopped the bot for Saturday and resumed it for Sunday), but has hit two profit targets since to be ahead. This is the sort of pattern I expected based on the modeling in the strategy details post where it seemed clear that stop losses would occur occasionally but the profit targets would be more frequent and would overcome the losses and then some.
Laying horses in the $2-$3 odds range in the ANZ Horse Bot using a 25c profit target per race and $100 stop loss
It’s a slow and steady system which is working well. Saturday was an absolute standout of a day, but I slept through most of it so I didn’t get to enjoy it as it happened, and woke to a rather nice increase in profit later.
The strategy which I ran through previously was working but required more patience than I was willing to provide to it. Effectively it was profitable however it would have some bad runs which would create a large drawdown, and a number of slightly-better-than-break-even days, and then the occasional standout period which would rocket it back up into profit. A good system if you have patience and a bankroll capable of withstanding some big drawdowns along the way. I wanted to try something different so I ran some numbers on a few days worth of Australia and New Zealand dog races to see if laying the third and fourth favourites at odds of under $8 would work with this staking system. The benefit of under $8 is that after a loss, the recovery cycle recovers into profit without needing further wins and can be in quite a good profit if the odds were quite low on the losing bet.
I had looked at something similar in my early attempts at this staking system but had included some higher odds runners and wasn’t taking anything under $5 as I figured anything paying less than that would be too much of a risk. The results then weren’t great and led to my further revisions into the strategies which I posted previously. A difference here is that I wouldn’t be even looking at odds over $8 and would take anything less, which means that at worst it only would need a strike rate of 83.33%. What I found from running the numbers was that the 3rd favourites under $8 could get me a strike rate of 86.52% and the 4th favourites 89.01%.
I decided to launch this strategy on Tuesday morning in the wee hours while the UK dogs were running. It took off like a rocket making $40 profit by 9am. After that the Australian day wasn’t great, and it has been up and down on the third favourites ever since, making profit overall but acting much like the rollercoaster ride of the previous systems on a shorter timeframe. The fourth favourites however have just been a steady upwards trajectory and have made $50 profit over the two and a half days.
It is worth noting that the third favourites have a lot of activity through the AU/NZ races and also through the UK races, with the evening session (UK time) of the UK dogs being the best performer so far. The fourth favourites get minimal activity through the Australian and New Zealand races, with one or two bets per meeting usually, but quite a bit of activity in the UK which makes sense as the smaller field sizes would lead to lower odds on a fourth favourite. The fourth favourites have less activity but more consistient results.
I have decided to run with the fourth favourites and stopped the third favourites.
I’m also working on a soccer laying strategy which has made a dollar in the day and a half it has been running. Weekends have much more activity than weekdays for soccer, so I’m hopeful that results will continue on their current path and I can share details with you next week.
Until then, greetings from myself and Pebbles and Shyley
In fact it should be possible to back and lay the same horse and come out a winner no matter who wins the race.
It’s March and the autumn racing carnival is just starting to appear, so it’s a good time for me to demonstrate a strategy which I have been using to generate slow and steady profits, and a derivative of that strategy which I have put through its paces in a spreadsheet, have confidence in, and have launched into production today.
I am using two bots for this, both of which I’m really just scratching the surface of their capabilities by using them in very specific ways. First up the ANZ Horse Bot is running my laying strategy. The settings I’m using in it are remarkably simple but have been honed over some time to get them to the point where I consider them mature and reliable. The bot itself though has an incredibly powerful stats engine which I’m not currently using, but can be used to find the runners which meet almost any form indicator or bit of statistical criteria you could think of, and then place bets on them per your instructions.
The other is the Fav Money Staking bot, which I’m using for the backing strategy. Again, I’m using quite simple settings, but the abilities of the bot go far beyond how I’m using it and I will be interested in exploring those options further after I have bedded-in the backing strategy to the point where I’m comfortable that it works in reality as well as it does on paper.
I have a video up today running through the details of my strategies, a bit of the thought process behind them, how they’re set up in the bots, plus a quick look at some of the more advanced functions of the bots.
But it’s also important to document the strategies here, so here we go.
The laying strategy
I have been running this strategy for a while now and it has been a consistent performer. It’s not going to set the world on fire but it is consistent and should scale up a bit from where I have been using it.
The premise is simple. In the odds range of 2.00 to 2.99, lay the horse with the lowest odds. Often this is the favourite in the race, sometimes it’s the second favourite. In some races there are no horses in this odds range in which case the race is skipped. Favourites tend to win about a third of the time overall in horse racing, with their strike rate generally improving at lower odds. In the 2.00 to 2.99 range, I find the lowest priced runner wins about 40% of the time, so lay betting gets a strike rate of 60%.
This strategy works with a Stop At A Winner staking plan. That is, you have a set profit target to reach per winning race, and you increase your staking after each loss so that you can reach your target and recover the losses in a single win. Then you stop increasing staking and start again with a new series of bets aiming for a target. Lay betting with SAW staking at this odds range has fairly gentle stake increases and so works quite well.
I have been running the system at a very moderate target of 25c per win (I think I got mixed up and said 50c in the video) across thoroughbred racing and harness racing, and have been seeing an average return of a bit over $3 per day. Saturdays tend to outperform everything which a much higher strike rate and profit rate, which makes sense given the quality of racing on a Saturday is generally higher and thus it’s harder for the favourites to defeat their competition.
This is the results graph from the last three weeks of bets
$69.39 profit in three weeks. Doesn’t sound like much when put like that, but considering this is off a profit target of 25c per winning race, I consider it quite a tidy little sum.
Most importantly, it grows consistently. And while some days are slightly tougher than others, the system hasn’t come close to reaching a stop loss and rarely has more than a few losses in a row which makes it quite safe in performing its loss recovery as per the Stop At A Winner Principle.
These are the settings I have been using.
(click image to enlarge)
They are fairly open in terms of which races qualify as I don’t see much point in reducing the volume of races when there are already plenty of races which don’t have a runner in my odds range and thus don’t qualify. As long as there is some liquidity in the market and there’s a horse meeting the odds criteria, I’m happy. If you wanted to take steps to reduce the volume of races and/or possibly increase the strike rate, increasing the mimiumum amount matched in the market or adjusting the number of runners in the race may achieve this. But I haven’t seen the need to do so and thus haven’t tested it.
One important thing in these settings is the inplay action is set to Take SP. This means that if the bet isn’t matched by the time the race jumps, the volume of unmatched bets with the Take SP are put through a calculation and get matched together at whatever prices the weight of the market dictates. It’s similar to the TAB dividends process. The odds you get from this are usually better than TAB odds but might be worse than you had been trying to get matched at, meaning if the race goes your way, you might not win quite as much as you hoped. But I consider that a better outcome than getting only partially matched (which can happen in rapidly moving markets) and barely getting any return at all. You might prefer to not have this process and to cancel an unmatched or partially unmatched bet at the jump, but I don’t recommend it.
The Backing Strategy
As I mentioned, the horses in this odds range are winning about 40% of the time. A 40% strike rate on bets in horse racing is a good achievement, but when the odds on offer for them is as low as this ($2.00 to $2.99) it’s actually a bit of a challenge to make a profit out of it. I had tried using the SAW staking method on backing these runners but found it was far too easy and frequent to have a losing streak of 8, 9, 10 races in a row and hit a stop loss, and it happened regularly enough the wins could not overcome the losses.
Still, 40% strike rate. There had to be something that would work.
Well, Steve and Michael who produced the ANZ Bot I’m using for the laying strategy do quite a bit of work on staking strategies and often then produce bots which take advantage of the unique features of each staking strategy. One strategy I had been looking at and thinking it might be a good fit for my 40% strike rate on short price runners conundrum is called Fav Money Staking because it is designed with the odds range of favourites in mind. It uses a more gentle staking progression than Stop At A Winner and doesn’t expect to turn a profit out of one winning race but rather out of a few in a series.
They have a bot to go with it. The Fav Money Staking bot. My only concern with it was that even if it was suitable for my specific horse selection criteria, as far as I could see it didn’t have the same option as the ANZ bot to pick “any selection” within an odds range and instead required you to choose a favourite ranking (ie. first favourite, second favourite, etc). This concerned me because it would require me to split my selection criteria in to two parts:
1. 1st favourite between 2.00 and 2.99
2. 2nd favourite between 2.00 and 2.99 if the first favourite is under 2.00.
This would result in the staking running separately on these strategies and potentially result in significant differences in outcomes to what I had modeled. Thankfully, it turns out my fears were unfounded.
I discovered while preparing the video that the “any selection” option is available in Fav Money Staking Bot. Whether it was there the whole time and it just wasn’t clear to me from the available documentation, or it was added in to a version at some stage, I don’t know. But it is quite a relief because it means I can run exactly the selection criteria in Fav Money Staking as I can in ANZ Bot, and thus my modeling should be valid.
The selection criteria is one thing. Wrangling the options of the Fav Money Staking system is another. Ultimately it is a fairly straight forward system in that you give it a starting stake, and it works out the sequence from that starting point, continuing until either the profit target or stop loss are met.
To model this and figure out how best to set this up, I took my lay results from the last three weeks, plotted them in Excel as if they were Back bets, then figured out the staking based on a 50c starting stake and plotted the results with various profit and stop loss settings.
At first I had no real success. Any combination I tried with the same selections as ANZ Bot just didn’t want to turn a consistent profit.
$5 profit target / $50 stop loss
It turned a profit, but hit stop loss three times and spent most of the three weeks circulating around the same spot. The pattern showed clearly that it would probably only break even in the long run.
So I took my observations about Saturday racing being tougher for backing winners and pulled all of the Saturday races out. With the same profit target and stop loss, things improved.
$135.63 profit, in fact. There was a stop loss in there, but it didn’t seem to do much damage.
Still, it did take the better part of a week to undo a stop loss, and that would be a concern in a period with a worse strike rate, so I kept fiddling.
$25 profit target / $50 stop loss / no Saturdays $263.58 profit. A much faster rate of increase along the way makes two stop losses look like blips. The faster rate of increase means that stop losses are minor inconveniences rather than generators of disappointing weekly results. Quite clearly by extending how long the staking progression has to run for, it resulted in the bigger stakes later on in the progression ultimately landing on enough winners to fly ahead.
I also tried a $20 target but it had very similar results to the $5 target
And at the other end of the scale, a 1c profit target, effectively meaning the staking progression should rarely ever need to go very far and thus limiting the risk of stop loss.
It avoided a stop loss even if one staking progression did go on for a bit and bring some drawdown briefly, but it also was quite slow compared to the $25 target and, if it had reached the stop loss, would probably have taken a couple weeks to dig out from it.
So after all of that modeling, I settled on a $25 profit target and set about setting up the bot accordingly.
These are the settings I am using.
(click image to enlarge)
Now, I set that up this morning and we haven’t quite reached today’s racing yet so I don’t have any real world results yet, and it is Friday today so while I anticipate a decent result today, tomorrow is Saturday and the system calls for no bets to be placed on Saturday racing, so I will have to stop the bot from placing bets tomorrow (either by pushing the stop button or deselecting the races…the stop button is the better idea, I just have to remember to start it again on Sunday morning). As such, I don’t expect to have useful results to show you until late next week at the earliest when a decent number of bets have gone through and I can see how the results compare to the modeling and, hopefully, be able to report it is all tickety-boo.
Combined strategies
I am continuing to run the lay strategy as it has proven itself to be a reliable workhorse for a while now. The upshot of running both the back and lay strategies simultaneously is that I will almost always be both backing and laying the same horse, meaning I will have a bet on for a specific horse to win and another bet for that same horse to lose. Sounds mad when you put it like that, but as the different strategies will be at different points in their staking plans and indeed are using different staking plans, they should both win in the long run across the multitude of races.
So effectively, it doesn’t matter who wins the race, the very fact that the race is being run is enough to generate a profit if these systems both work according to plan.
If you’re interested in trying something similar, or maybe entirely different strategies, the two bots I’m using and plenty of documentation about them are available at these links: Fav Money Staking Bot ANZ Both
I can’t wait to see how these strategies perform together. It’s going to be fascinating. I look forward to bringing you some results soon.
It would be fair to say that last week’s revision to the selection criteria for the greyhounds in the LayPro88 staking system didn’t quite go according to plan. The third favourites improved quite a bit and were roughly breaking even throughout the week, ending very slightly down. Good news but still room for improvement. The fourth favourites though started off badly (as seen in my day trip to Goulburn) and stayed there, sitting roughly level most days but with a downward trend overall.
Breaking it down into odds ranges tells a bit of a story
(HTML table formatting is a pain, it’s much easier to just take a screenshot out of Excel rather than paste the data in and fix up the HTML formatting, especially on a table this size).
It is notable that the strike rate on the higher odds for the fourth favourites dropped off by about 5-10 percentage points. This explains most of the trouble. With five bets in the recovery cycle following a loss, it is not possible to fully recoup a loss on anything paying over $8 without more winning bets after the recovery cycle. As such, the higher the odds, the higher the required strike rate, but at the same time the higher odds tend to improve the strike rate and the systems don’t turn a profit without something in the higher odds ranges. The dropoff in strike rate at the top end of the odds was costly. There will always be some variability in strike rate in any form of racing and it’s possible this was just a bad week, but it’s important to be able to handle a bad week and the performance in this week was worse than I would like to see from a bad week. Eliminating losses entirely isn’t possible, however reducing the damage they do is the aim so that the profitable days can far outweigh the bad ones should be achievable.
Based on this data, I have reduced the top end of the odds range from $15 to $13 for both the third and fourth favourites. The fourth favourites have a decent chunk of activity in that range, and losses are very costly up there. When the strike rate drops at that end, it is virtually impossible for the rest of the system to compensate.
I have made the same change on the third favourites. While the losses at that end of the third favourites remain rare, they are still costly. If things go well I might reconsider allowing the third favourites up to $14, but I’m not sure that the greatly increased strike rate up there is worth the risk of losses.
On the low end of the scale, the third favourites remain at a $6 odds minimum. This end continues to perform well despite a lower than desired strike rate as a five bet recovery cycle recoups more than the lost amount. The strike rate for the fourth favourites below $7 was poor when it was in use previously, and the handful of bets which got placed at the lower odds due to market flucturations at bet placement time remained at about the same strike rate, so the $7 minimum seems to be fine there.
So, starting today the new odds ranges have been set as:
3rd favourites: $6-$13
4th favourites: $7-$13
And another week begins…
Incidentally, I’m running LayGreyBot in a virtual machine along with a few other bots. While the bots themselves don’t use much memory, their memory usage does increase over time but tends to level off after a few days. Windows 10, even with as much of the unnecessary preinstalled bloatware disabled as possible, tends to take up a bit of memory and doesn’t always handle that expansion of memory usage from the bots very well. The host machine for the virtual machine is used for other purposes intermittently and doesn’t have a huge amount of RAM to begin with, so I originally only assigned 2GB of RAM to the Windows VM, which was fine with one bot when I started out, but with multiple bots running it started to cause unexpected application crashes and the odd full Windows reboot. 3GB was better but not great, I’ve found 4GB is fine for running multiple bots. My advice for Windows 10 is allocate 1GB for Windows and 500MB per bot, then allocate an extra 1GB if running more than one bot. I haven’t tested Windows 11 but wouldn’t expect it to be any more efficient.
At some stage I’ll probably look at moving the bots onto another machine with more resources. Whether that’s the existing VM moved on to more generous hardware, or whether I build a new environment as either a physical standalone machine or a new VM, I haven’t decided yet.
On Friday I ventured out on the bike up to the Goulburn greyhounds and had an enjoyable day following my automated systems as they bet on the races…well, until the weather turned anyway!
If you’re interested in more details of the strategies in use, I have a couple blog posts detailing that here and here.
The video is a tad on the long side so I have broken it up into chapters for your convenience.
0:00 Introduction
2:04 Staking systems
8:29 The auto-betting software
11:45 Race 1
16:40 Race 2
20:23 Advantages of automated betting over manual betting
21:53 Race 3
24:41 The importance of appropriate staking
26:32 Race 4
31:40 Race 5
35:57 The storm hits Goulburn
36:38 Race 6
42:37 Summary
It might sound strange to say that I have been waiting eagerly for a losing period on the LayPro88 greyhound system, but it has come and I’m actually very delighted about it. If you’re not sure what I’m talking about, my previous post outlining and explaining the system is here.
With any betting system, I always want to see a bad day in action. It’s great to have good days with nice profits, but it is inevitable that betting systems will have bad days, so with any betting system I always want to see it have a bad period so that I can see where the real weaknesses are. It’s impossible to completely prevent days with losses from occurring and one should always be prepared for that, but data gathered on a bad run is invaluable in showing where the weaknesses are and what can be adjusted to reduce the frequency with which bad days will occur. Ultimately the aim is to reduce the bad days so that, although they will result in a loss, that loss is inconsequential when considered alongside all of the good days.
When I set up a system, I always wait for the first bad period to come along and then I analyse the data from those days and determine whether
a) there is a weakness which can be removed, or
b) the system as a whole is the issue and needs serious revision
The greyhound system has been going so well that it has taken a few weeks to get there but now that the bad period has visited, I’m delighted to say it fits category A.
You may recall I have been running with two similar systems running concurrently.
1. Lay the fourth favourite if it is paying odds of $5.00 to $15.00 and is paying at least $2.00 more than the favourite.
2. Lay the third favourite if it is paying odds of $5.00 to $9.00 and is paying at least $1.50 more than the favourite.
The fourth favourites have been steaming along making nice profits, so I’m not too concerned about them but took the opportunity to analyse them a bit anyway. The third favourites on the other hand have been troublesome. It stands to reason that the third favourites would win more often than fourth favourites and would pose more trouble with this system which is betting that they will lose, so this wasn’t unexpected, but was a good opportunity for analysis and revision when the trouble finally provided enough data.
The third favourites have been struggling for a few days. Some periods of profit but mostly small losses slowly accumulating. You may recall I originally set the upper odds limit at $9 because the recovery cycles can fully recover any losses at this amount without needing extra wins. Unfortunately this odds range was not producing a high enough strike rate to always get through the recovery cycles, so the stake sizes increased slowly over a few days until the reached the stop loss limits I had set (the stakes had grown from the $0.50 starting point to over $6). At this point the LayGreyBot accepted the loss, gave up on trying to recover it, and started afresh with a new cycle at the original stake size (it can also be set to stop completely).
Over the course of four days, 17/2 to 20/2, the 3rd favourites system placed 652 bets and had 547 wins, a strike rate of 83.90%. Well short of the 88% generally required to make consistent profits in a LayPro88 system, and short even of the 84.5% the third favourites have previously needed to return a break even result.
I decided the best option was to perform a breakdown of this period by odds range and see if some odds ranges were pulling the whole system down. It did not disappoint. Note that even though the systems is set to $5.00-$9.00, occasionally it will pick up a runner just outside of this range due to variability in the market in the seconds leading up to the race.
Odds range – third favourites
Bets
Wins
S/R
4-4.99
5
4
80.00%
5-5.99
196
157
80.10%
6-6.99
229
197
86.03%
7-7.99
121
104
85.95%
8-8.99
82
67
81.71%
9+
18
17
94.44%
Total
652
547
83.90%
6+ odds only
451
386
85.59%
Clearly anything under $6 is a problem. The $8 range was strangely problematic too but with a relatively low volume of bets, so not worth removing from the system given the 9+ range which was performing well would also need to be removed in order to do so.
Out of interest I decided to see how third favourites at higher odds ranges were performing. $9-$15 seemed like a good range given the fourth favourites are betting up to $15 odds. As the bot has not been looking at these ranges, I didn’t have data to hand which I could simply analyse in a spreadsheet, so instead I had to go through Betfair’s results pages and check manually race-by-race. I only went back as far as the 18th but got more than enough data to reach a conclusion.
Date
Total eligible
Total wins
S/R
18-Feb
17
17
100.00%
19-Feb
20
20
100.00%
20-Feb
35
29
82.86%
Total
72
66
91.67%
If added to existing odds range
724
613
84.67%
If added to 6+ odds range
523
452
86.42%
The higher odds, while low in volume, add a bit to the strike rate, and even half a percentage point on the strike rate really can make a big difference with this system. The solution based on the data is obvious. I have adjusted the odds range for the third favourites from the previous $5-$9 range to a new $6-$15.
And on the fourth favourites I ran a similar analysis and received some interesting results which also led to a change.
Odds range 4th favourites
Bets
Wins
S/R
4-4.99
0
0
N/A
5-5.99
13
8
61.54%
6-6.99
118
97
82.20%
7-7.99
158
136
86.08%
8-8.99
159
142
89.31%
9-9.99
77
71
92.21%
10-10.99
58
51
87.93%
11-11.99
86
82
95.35%
12-12.99
65
62
95.38%
13-13.99
41
41
100.00%
14-14.99
45
39
86.67%
15-15.99
19
19
100.00%
16
0
0
N/A
Total bets
839
748
89.15%
7+
708
643
90.82%
Anything under $7 seemed to be performing quite poorly here, and although the system was running profitably, an extra percentage point or so on the strike rate doesn’t hurt at all (quite the opposite actually!!) so I have revised the odds range for the fourth favourites from the previous $5-$15 to a new range of $7-$15.
And now the new ranges are underway and the monitoring continues.
Or $206.65 in seven days…but I’ve only compiled full stats for the four and a half days
I’ve been working on automated betting strategies for greyhound racing for a little while now, and when I’m working on it I tend to fiddle with the settings in response to how things do or don’t work, and also towards what I think might or should work. The more I fiddle with the settings, the harder it can be to determine how well (or not) it’s working as these things need to be given time to work through multiple days and see how they go with different tracks, qualities of fields, weather conditions, etc. Ideally an automated system should, once set up, be able to handle most variations in conditions with minimal drawdown and fairly consistent returns. It’s simply not possible to always win, so it’s also essential that an automated system has returns on good days which outweigh losses on bad days and ability to pull up the handbrake if things get too bad without blowing the bank, especially given an automated system will go unmonitored for periods of time and a lot can happen in a short space of time.
I’ve been pretty happy with the results of the greyhound systems I’ve been using for a bit of time now and decided it was a good time to put a moratorium on adjustments and just let them run and prove themselves. So late on Monday afternoon I did just that and let the systems run through until Saturday morning. The results impressed me more than I expected.
I had three strategies running on auto-pilot.
1. Lay the 4th favourite (that is, bet that it will not win) if it is paying odds of between $5 and $15, and is at least $2 more than the favourite, on Australia and UK races. It made $67.22 profit
2. Lay the 3rd favourite (that is, bet that it will not win) if it is paying odds of between $5 and $9, and is at least $1.50 more than the favourite, on Australia and UK races. It made $63.51 profit
3. Back the favourite if it is paying odds of between $3 and $5, on Australia and New Zealand races. It made $6.93 profit
For the benefit of anyone unfamiliar with the concept of lay betting, this is a feature available on the Betfair exchange and is effectively the opposite of backing a horse/dog/whatever to win an event. Sounds simple, certainly easier than picking the winner, and it is, but at significantly reduced odds so it requires some careful staking strategies to make it worthwhile. If you’re familiar with the concept of betting on something to win, it’s easy to wrap your head around betting on something to not win.
Suppose you bet on a horse to win a race. You bet $1 and it is paying odds of $7.50. If the horse loses, you lose the $1 that you bet, but if the horse wins you receive your dollar back plus $6.50, meaning you’ve turned your $1 into $7.50. For the person on the other end of the bet, it is the opposite. If you’re betting with the TAB or a bookmaker, then the TAB or bookmaker is the party with the lay bet, but in the case of Betfair you are betting against other users of the exchange. So in the example we just went through, for the person with the lay bet, if the horse loses they win the $1 that was bet on it, however if the horse wins, they lose $6.50 which goes to the person with the successful back bet.
In my case for the lay bets I am using a staking method called LayPro88. The 88 here refers to the magical percentage of 88% strike rate which is where this method starts to turn a profit. So yes, in order for this method to work, the dog I am laying has to lose 88% of the time or more in order for me to turn a profit. I’ve seen it turn a profit at a lower strike rate, but 88% is a good guide number, and anything above that really starts to turn some fantastic profits.
The system is quite simple to follow. To start off, I lay the dog with a stake (the bet from the person backing the dog) of 50 cents (it doesn’t matter what the starting stake is, this is just the amount I have been using to go with the amount of bankroll which I attributed to each strategy, which was $100). If the dog wins, great, I got 50 cents profit, go again. But if the dog wins, then I have a loss and I add my original stake to the most recent stake, so my next stake will be $1. If that wins, then the next stake is reduced by 20% of the original stake. This process continues on an ongoing basis, but the stake can never go below the original stake.
So for example, a sequence of bets could look something like this:
1. 50c, win
2. 50c, win
3. 50c, loss
4. $1.00, win
5. $0.90, loss
6. $1.40, win
7. $1.30, win
8. $1.20, win
etc
For the back bets I am using a different system which requires a bit more calculation. It’s called “Stop At A Winner”. It works on the basis of trying to get a specific profit out of a race. In this example I was aiming for 12 cents per win. It is probably a bit on the conservative side and I could increase it a bit, but I am wary of the drawdown which can occur with this method (as could be seen multiple times in the graphs above) if there’s a bunch of races in a row where my selection does not win.
Effectively the way it works is this: In the first race the aim is for 12c of profit and so a bet is calculated based on the odds available to make that profit. Eg, if the dog is paying $3, the bet would be $0.06 (a 6c bet, if won, would get 18c back, 6c being the original bet and 12c being profit). If the bet wins, great, put that aside and start again. But if the bet loses, the 6c loss is added to the profit target so effectively in the next race we’re trying to make 12c profit plus recoup the previous loss. And this goes on with the stake size increasing until a winner is found. This tends to have a good steady growth, but the drawdown requires careful risk management and so reduces the bet size and profit a fair bit.
You could use these methods manually if you wanted, but in order for them to work particularly well it is important that the betting market is well-formed by plenty of bets having been placed so that the odds are a good reflection of what punters believe the chances of each runner are, so I have these bets being placed about 30 seconds before the scheduled start time of the race. This doesn’t leave a lot of time to place the bets manually, and while the LayPro88 method wouldn’t be too hard to do as you know exactly what stake size you need based on the result of the previous race, the Stop At A Winner method requires calculating and being able to get your bet on before the odds change (and the odds do move very quickly on a lot of races in the seconds before the race starts).
There is also the question of time. How much time do you want to spend on this? A tradeoff in my view is that an automated strategy should always have smaller bets than you would place manually to avoid something going wrong with the method and a terribly calamitous loss occurring without you seeing it, but at the same time an automated strategy can bet at all hours of the day and night so small bets can add up to quite sizable profit quite easily.
When you consider that Australian and New Zealand greyhound racing often starts around 11am eastern Australian time (earlier if NZ has a morning meeting and on Saturdays when Australian greyhounds start early to avoid racing in the afternoon during the premier thoroughbred horse racing) and run through until midnight or later, and UK dogs start mid-evening Australian time and run through until around 6-8am depending on the time of year, this is where an automated system can excel by betting 20+ hours of the day for you while you get on with life. In the examples in this post, the bots were doing the work while I was at work, or asleep, or doing grocery shopping, or playing with Pebbles and Shyley, or taking the motorbike out for a ride, or one of the million and one other tasks which crop up in life.
I’m using two automated tools for this. One for the LayPro88 method and one for the Stop At A Winner method.
This is the main screen where you can see an awful lot of settings down the side which allows you to setup whatever strategy you want, and make it as selective or as open as you want. For example down the bottom you can go as far as selecting what types and grades of races you’re willing to bet on, you can choose which boxes you’ll accept the dog to start from, further up you can set things such as the minimum and maximum race distances you’ll accept, even how much other people have to have already bet on the race before you’ll consider it. For my method I have the settings pretty wide open so as long as there is some money in the market and the 4th (or 3rd) favourite meet my odds criteria, I’m pretty much happy to place the bet. I have a couple safeguard in place to ensure bet sizes can’t get too big and the market has some real liquidity to it, but that’s something I might dive into another time. It’s probably a bit too in-depth for this post. On the other side you can select or deselect individual races or entire meetings if you want, but I usually leave them all enabled.
The bot allows you to run up to three strategies simultaneously, so I have my fourth favourite strategy in “Lay 1” and my third favourite strategy in “Lay 2” (why not the other way around? Because I started with the fourth favourites first and added the third favourites later on after monitoring them for a while).
In addition, and this is where the bot really improves things over betting manually, it can run multiple races at once in the various tabs at the top of the screen. This is fantastic when you have multiple races occurring at about the same time, or a result from one race takes longer than usual because of a photo finish and the next race is due to start before the result of the previous race is known, or even just for delays in getting races started where you have placed a bet shortly before the advertised start time of a race, but it doesn’t get underway on time and you have other races to get to, something which happens quite often, especially with some of the UK meetings which can end up running an hour or more late sometimes.
Earlier on I mentioned that the stake size for this method depends on the result of the previous race, so each tab operates independently of the other tabs. Eg. a win or loss in tab 1 only affects the stake size of the next race in tab 1, likewise a win or loss in tab 2 only affects the stake size of the next race in tab 2, etc. The general recommendation from many people and even from Steve and Michael in their documentation for this and other similar bots is to only use a single tab for the LayPro88 method. This is good advice to avoid having multiple tabs hit a losing streak and all go down a fair way and need to recover through the process of getting more wins. But at the same time, knowing how many greyhound races there are and how a delay in one race can cause you to miss a whole heap more races, I actually run with all five tabs active.
Usually through the Australian races I only ever see activity in the first two tabs, but the British greyhounds tend to stretch out a bit more across the tabs due to the fact that British race results are slower to come through than Australian results, there’s a higher volume of races per hour in the UK most of the time, and often at least one race meeting is running behind schedule.
I wouldn’t recommend running multiple tabs until you’re comfortable with your strike rate percentage being around that 88% mark or better, but if you are using multiple tabs, my method for managing the recovery which some tab will inevitably need is that if one tab is comfortably in profit while another tab needs a greater volume of races through it to speed up recovery, I will temporarily disable the profitable tab to funnel more races to the tab which had been struggling. If the method is able to maintain the 88%+ strike rate over the long run, a bad day with a poor strike rate should be offset pretty soon by a day with a better strike rate, so it makes sense to feed a better day into the tab(s) which need it most.
For example, you saw the ups and downs of the overall results in the graph earlier, here are my results over the course of the days broken down by day and strategy. Note for this the first Australian day started about half way through the day’s racing, and I have treated the UK days as being entirely on the date on which they started in Australia, so for example the 12/Feb UK day ran from about 10pm 12/2 AEDT to 9am 13/2 AEDT but I have considered it all to be 12/2 for the sake of consistency in data reporting.
12-Feb
13-Feb
14-Feb
15-Feb
16-Feb
Total
3rd fav AU profit
-$2.00
-$24.41
$22.30
-$46.42
$52.77
$2.24
3rd fav AU S/R
80.000%
79.688%
86.842%
80.952%
92.424%
84.735%
3rd fav UK profit
$32.42
$41.38
-$35.79
$14.89
$8.37
$61.27
3rd fav UK S/R
90.476%
89.583%
81.928%
84.884%
84.821%
86.515%
4th fav AU profit
$13.55
$21.81
$34.20
-$4.46
$8.47
$73.57
4th fav AU S/R
97.297%
89.610%
96.471%
89.216%
90.217%
91.858%
4th fav UK profit
$33.06
-$48.42
$18.75
$21.73
-$31.47
-$6.35
4th fav UK S/R
93.162%
85.950%
89.764%
89.565%
85.294%
88.636%
Day total profit
$77.03
-$9.64
$39.46
-$14.26
$38.14
$130.73
Day total S/R
93.426%
86.872%
89.218%
86.563%
87.685%
88.018%
Monitoring this intermittently as it was going, I really thought the third favourites were struggling more than the stats demonstrate. They did have a lower strike rate than the fourth favourites, which is to be expected as a third favourite is by definition more likely to win than a fourth favourite, but managed to get some decent profits despite generally being below the magical 88% mark. I put this down to the fact that the upper odds limit I set on the third favourite strategy of $9 is the maximum amount for a loss to be fully recovered by the recovery cycle, and thus with most of the bets in this category having lower odds, any loss was well-and-truly recovered and then some by subsequent wins.
The fourth favourites did better in Australian than in the UK, probably due to smaller field sizes in the UK (they’re a maximum of six dogs per race whereas Australia runs up to eight per race) and the UK appears to have come in with a slight loss despite hitting 88%, but this is at least partially because the last couple of hours of the UK races had some losses which weren’t fully recovered at the point where I took the statistics.
It’s worth noting that the odds range on the fourth favourites extending out to $15 even though a loss on anything over $9 can’t be fully recovered in a recovery cycle (it would require more winners after returning to the original stake size) is done for a reason. Firstly, many fourth favourites have odds greater than $9 so you miss a lot of races if you limit the odds range, but more importantly it helps to keep the strike rate up. I did try reigning in the odds range to $5-$10 but it shaved about two percentage points off the strike rate and made the strategy only just break even overall. It stands to reason that the higher odds runners win less often and so the risk of bigger losses along the way is worthwhile when it also brings a better strike rate and better winning runs. At some stage I will analyse the strike rate of the fourth favourites in bands of the odds range and see if the odds range can be optimised a touch, but for now it’s working quite well.
The LayPro88 method has a lot of ups and downs and requires patience, but as long as the 88% or thereabouts can be maintained in the long run, it comes out quite nicely ahead. You can see that the overall figure was indeed just slightly over 88% strike rate.
There was a point in this where tab 1 was quite happily running ahead in profit while tab 2 had suffered a few losses, so I did disable tab 1 for a couple days so as to funnel the bulk of the races into tab 2. Apart from that, the system ran by itself with no intervention or input from me. In fact, four days and a half of racing, 1,811 bets placed in total, and I probably watched about 30 races in total.
One good thing about running the third and fourth favourite strategies simultaneously is that in many races, bot strategies will place a bet, and at least one of them has to win, usually both do. This helps to slightly reduce any losses accumulated in any given race.
An example from a UK race this morning at The Valley.
The system placed bets on two runners, one was third favourite and the other fourth favourite. The third favourites had been through a slightly rough patch so the stake size was up at $2.40 and the potential loss $13.44, while the fourth favourites were doing well so it was at the base stake size of 50c with a potential loss of $4.10.
Due to the fact that at least one of these runners had to lose the race and be a successful lay bet, the potential losses on the race were not as high as the individual bets showed. If they both lost the race, it would be a profit of $2.90. If the third favourite won, it would be a loss of $12.94, not $13.44. If the fourth favourite won, it would be a loss of $1.70, not $4.10.
As it happened, the favourite won the race.
So both lay bets were winners.
I’ll probably go into some more detail of the interface of the LayGreyBot at a later stage, but it’s very nice to have an overall results tab showing you all of the results from your bets as they happen.
(click image to enlarge)
In addition to being able to go into each tab and see not only its results but exactly what it is doing at this moment and how its profit/loss stats look. In this picture, tab 1 has a bet in progress in Traralgon and some recovery to do on the third favourites while the fourth favourites are steaming ahead.
(click image to enlarge)
I’ve spent a lot of time here focusing on the LayPro88 strategy and the Lay Grey Bot but of course I also have the Stop At A Winner strategy working on the Australian and New Zealand greyhounds. This, as I mentioned, is backing the favourite in the race if it is at odds of between $3.00 and $5.00.
Functionally it looks and works very similarly to the LayPro bot, but with different settings options as it is designed to work with the Stop At A Winner staking method.
The results graph shows a bit of drawdown along the way, but the daily summary shows the consistency of the method.
Day
Bets
Bets won
S/R
P/L
Running total
12-Feb
23
3
13.043%
$0.17
$0.17
13-Feb
29
8
27.586%
$0.93
$1.10
14-Feb
48
13
27.083%
$2.04
$3.14
15-Feb
44
10
22.727%
$0.96
$4.10
16-Feb
43
16
37.209%
$2.83
$6.93
Total
187
50
26.738%
When you consider that this bot had about a tenth of the number of bets as the LayPro system and a much lower strike rate around 26% compared to 88%, the fact it managed to make $6.93 or about a 20th of the LayPro system’s profit is actually quite a remarkable and respectable result.
I expect I will take a closer look at this bot in a future post, but for now suffice to say I am quite happy with the bot running the settings I have described and the quite consistent profit it manages to generate. Sure, it’s a smaller profit than the other method, but this is largely down to my risk-averse settings and I could turn it up a bit if I wanted to.
A question which probably arises from all of this is why am I working with Australia and the UK on the LayPro, and Australia and New Zealand on the Stop At A Winner. Why not all of the countries in both systems?
The statistics from a bit earlier on where the 4th favourite runners won races more frequently in the UK than in Australia helps to tell the story. I did originally try a few variations of the Stop At A Winner method on UK greyhound races but found it to be unreliable with lengthy runs of losses which increased staking beyond what I was comfortable with and hit my stop loss limits regularly. It was interesting in a way as it would have a run of really good days where the profit shot up, but then lose it all and more a couple times in a row in subsequent days.
This is one of the challenges with testing automated methods is getting a feel for when a system is producing results which are too good, and are probably a sign that a very poor period is ahead. In the case of the UK greyhounds, the narrower tracks tend to result in more bumping which makes it a tad harder for favourites to win, certainly enough of a difference to make the method not work well at all. From my testing, I believe it could be made to work if I was very selective about the tracks and distances which the system bets on, but I’m not convinced the small increase in profit it worth the time to work out which venues and distances would be a good fit. Perhaps that’s something you might be interested in pursuing.
As for the LayPro system not betting on New Zealand, well I would like to bet this method on New Zealand greyhound races and my perusal of race results indicates it should work very well, which is also backed up by the fact that the Stop At A Winner system does particularly well on favourites in New Zealand, but it seems that something in the way Betfair is describing the New Zealand races has changed and so the bot is having trouble working out if the races qualify or not. I have reported the issue and hope a fix will be delivered soon.
I started this post telling you how I made $137.66 on the greyhounds on autopilot in four and a half days, and that over seven days the total has reached $206.65, but I haven’t explained that yet. I have, however, shown you these figures throughout the post. You see, when I took the statistics on Saturday morning around 8am, I reset the bots, so they started from scratch with an entirely new set of statistics and staking. The screenshots above tell the story in the big boxes in the middle top of the main bot screens. The screenshots were taken this afternoon around 4pm, pretty much exactly seven days since I set the bots and left them largely to their own devices.
LayPro88 (third and fourth favourites combined) first period $130.73
Stop At A Winner first period $6.93
LayPro88 second period fourth favourites $120.57
LayPro88 second period third favourites $55.23 loss
Stop At A Winner second period $3.65
Total: $206.65
Later this week I intend to put together a live demonstration of how these methods and bots work, not as a live stream but in video form showing the bots in action as some races go by, which I think will help to explain the system a bit better and provide a better feel for how the bots work and how I work with them. It should certainly be better than me posting another dozen or so screenshots and writing another slab of text like this one.
Of course if you’re interested in following my footsteps or exploring the methods and bots for yourself, here are the links again to the homepages of the bots where you will find a lot more information about how they work and even a few videos. Lay Grey Bot Stop At A Winner Greyhounds Deluxe
I also have some strategies on the go for horse racing and soccer, but they will keep for another day. At the moment my main focus is on the greyhounds where I have been seeing some significant positive results after a decent amount of time and effort testing strategies, and I’m looking forward to showing you the strategies in action soon.
